What is mean by 'cross-validation with non-unique 'x' values seems doubtful' in Smoothing spline in R?

Question

I am using smoothing spline function in R but I get a warning message saying:

  > boneMaleSmooth = smooth.spline( bone[males,"age"], bone[males,"spnbmd"], cv=TRUE)
    Warning message:
    In smooth.spline(bone[males, "age"], bone[males, "spnbmd"], cv = TRUE) :
    cross-validation with non-unique 'x' values seems doubtful

  > boneFemaleSmooth = smooth.spline( bone[females,"age"], bone[females,"spnbmd"], cv=TRUE)
    Warning message:
    In smooth.spline(bone[females, "age"], bone[females, "spnbmd"],  :
    cross-validation with non-unique 'x' values seems doubtful

I read it somewhere saying it does not matter much. But I am not sure what caused it. Hope someone can help me.

Answer 1

This means you have some data points that share an x-value. It shouldn't make too much of a difference to the practical result you get in most cases. However, using cv=F is a better way to do things with such data.

The reason behind it is in the smooth.spline code, it works out the total number of x-values:

n <- length(x)

And then it starts to process your data (I've included comments):

xx <- round((x - mean(x))/tol) #normalise the data
nd <- !duplicated(xx) #get rid of duplicated x values! THIS IS PART OF THE WARNING
ux <- sort(x[nd]) #Sort the data
nx <- length(ux) $Get the length of processed data to work with

Then, later, the warning is thrown if nx < n:

if (CV && nx < n) #CV is based on the input parameter `cv`
        warning("cross-validation with non-unique 'x' values seems doubtful")